In the last 16 years, more than a dozen gene-directed enzyme prodrug therapies for cancer treatment have been evaluated in preclinical studies. However, only few of them have evolved to the stage of clinical trial. This review assesses current knowledge in the area of cancer gene therapy, emphasizing cytochrome P450 (CYP)-based prodrug activation systems. This approach is intuitively highly suitable for the treatment of cancers, since several major anticancer drugs are activated by liver CYP enzymes. Important features of this strategy include: 1) use of human CYP genes to avoid immune complications that may hamper expression of therapeutic genes of non-human origin and thereby inhibit prodrug activation, 2) use of well established and clinically effective anticancer prodrugs, 3) strong bystander cytotoxic effect seen with all liver-activated CYP prodrugs, 4) the potential to inhibit liver CYP activity and expression to increase the bioavailability of prodrugs for CYP-transduced tumors, 5) possible extension to many CYP enzymes and their potential anticancer prodrug substrates, and 6) it can be used to arm therapeutic conditionally replicating viruses. Historically, this strategy utilized CYP 2B1 to activate oxazaphosphorines. It is now becoming clear that the repertoire of prodrugs is expandable and that CYP gene candidates are not limited to naturally occurring CYP genes, but may also encompass engineered CYP enzymes, improved by site directed mutagenesis or other approaches. Encouraging results from a recent phase I/II clinical trial that have implemented this strategy, as well as emerging problems related to gene delivery are discussed in this review.